Moser Proposition Al Knowledge Philosophical Studies 1987

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/288834422
Propositional Knowledge
Article in Philosophical Studies · January 1987
CITATIONS READS
2 3,374
1 author:
Paul K Moser
Loyola University Chicago
196 PUBLICATIONS 803 CITATIONS
SEE PROFILE
All content following this page was uploaded by Paul K Moser on 01 January 2016.
The user has requested enhancement of the downloaded file.

Propositional Knowledge
Author(s): Paul K. Moser
Source: Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition
, Vol. 52, No. 1 (Jul., 1987), pp. 91-114
Published by: Springer
Stable URL: http://www.jstor.org/stable/4319906
Accessed: 31-12-2015 22:27 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
Springer is collaborating with JSTOR to digitize, preserve and extend access to Philosophical Studies: An International
Journal for Philosophy in the Analytic Tradition.
http://www.jstor.org
This content downloaded from 147.126.1.145 on Thu, 31 Dec 2015 22:27:29 UTC
All use subject to JSTOR Terms and Conditions
PAUL K. MOSER
PROPOSITIONAL KNOWLEDGE
(Received 18 August, 1986)
Propositional knowledge is knowledge that some proposition is true.

It thus contrasts with knowledge-howand perhaps with knowledge-
who and knowledge-which.Since the time of Plato's Meno, at least,
philosophershave been vexed by the question of what exactly propo-
sitional knowledge is. The most troubling question is: What are the
logically necessaryand sufficientconditions for one's having proposi-
tional knowledge?Also since Plato's time, it has been widely held by
philosophers that propositional knowledge requires justified true
belief. Dissenting philosophers there have been, but the topic of
serious dissension has recently lain elsewhere: with the issue of the
needed fourthcondition for one's having propositionalknowledge.
According to the standard analysis of knowledge, suggested by
Plato and Kant among others, if one has a justified true belief that a
proposition is true, then one knows that that proposition is true.
However, since Edmund Gettier's celebrated counterexamples of
1963, philosophers generally have held that the standard analysis
needs modification.' Although there is, of course, no widespread
agreement on exactly what modification is needed, there are a few
prominent proposals.One sort of proposed modificationrequiresthat
the justification appropriate to knowledge be "undefeated" in the
sense, roughly speaking, that some appropriate subjunctive con-
ditional concerning genuine defeaters of justification be true of it.2
Another noteworthy modification requiresthat the justification for a
true belief qualifying as knowledge not depend on any falsehood.3
And still another modification proposes that the justification
appropriateto knowledge must admit of an "epistemic explanation"
that does not involve any falsehood.4
Serious objectionsto the prominentvariantsof the firsttwo sorts of
modification are too well-known to review here. The third sort of
PhilosophicalStudies52 (1987) 91-114.

? 1987 by D. Reidel PublishingCompany.
92 PAUL K. MOSER
modification is relatively new, and so will be considered, but found

wanting, in the opening sections of this article. Yet shunning the
growingdespairabout the prospect of successfullyanalyzing proposi-
tional knowledge, I shall argue for a surprisinglysimple and intuitive
addition to the justification condition for knowledge that blocks the
various counterexamples inspired by Gettier. The analysis based on
this addition might seem initially to be too straightforwardto be true,
but it will be shown below to withstand the most difficult Gettier-
inspired counterexamples.Overall, then, this article argues for a new
analysis of propositional knowledge that avoids the defects of its pre-
decessorsand solves the Gettier problem.
Robert Shope's recent analysis of propositional knowledge relies on

the assumption that such knowledge requires a certain kind of
explanation of the knower's actual justification. Shope's basic
proposal is that what is central to avoiding Gettier-style counter-
examples, in an analysis of knowledge,is the recognitionthat in such
examples falsehoodsplay certain roles in relation to a person's actual
justification. More specifically, the proposal is that in such examples
falsehoods are included in epistemic explanations of a person's
justifiedbeliefs and of variousjustifiedpropositionswhich are relevant
to the person'sepistemic situation, but which are not believed by that
person. An epistemic explanation, roughlyspeaking,is a set of propo-
sitions explaining why some propositionis justified(p. 208; cf. pp. 86,
94).5 In a Gettier-style counterexample, according to Shope's diag-
nosis, there is an ineliminable epistemic explanation that contains a
false proposition, and conversely, in a case of propositional knowl-
edge, the relevant belief is justified through its connection with a
sequence of epistemic explanations that do not involve falsehoods.
Following Shope, let us call a sequence of epistemic explanations a
'justification-explainingchain' (a 'JEC'for short),and let us grantthat
the members of a JEC must themselves be justified, if there is to be
genuine knowledge.
We can further clarify Shope's notion of a JEC by characterizing
the members of a JEC. The first member, ml, of a JEC related to a
PROPOSITIONAL KNOWLEDGE 93
proposition, p, is just a true conjunction of the form: '1f and that

makes the proposition that p justified'. In that conjunction, the prop-
osition thatfj describes something sufficient to make the proposition
that p justified. If ml has a successor,M2,in the relevantJEC, M2will
be a true conjunction of the following form: 'f2 and that makes mlI
justified' (where the proposition thatf2 describessomething sufficient
to make ml justified).A similar point can be made for furthersucces-
sors in the relevant JEC. The basic idea of Shope's diagnosis, then, is
that by constructing such a JEC for a Gettier-style counterexample,
we can expose false propositions, and thereby account for the lack of
genuine knowledge.
For purposes of illustration and evaluation, let us consider how
Shope's diagnosis would treat the following difficult Gettier-style
counterexample:6
(I) Suppose a person, S, knows the following true proposition, m:
Mr. Jones, whom S has always found to be reliable and whom
S has no good reason to distrust at present, has told S, his
office-mate, that p: He, Jones, owns a Ford. Suppose also that
Jones has told S that p only because of the state of hypnosis
Jones is in, and that p is true only because, unknown to himself,
Jones has won a Ford in a lottery since entering the state of
hypnosis. And suppose further that S deduces from m its
existential generalization, q: There is someone, whom S has
always found to be reliable and whom S has no good reason to
distrustat present,who has told S, his office-mate,that he owns
a Ford. S, then, knows q to be true, since he has correctly
deduced q from m, which he also knows to be true. But suppose
also that on the basis of his knowledge that q, S believes r:
Someone in the office owns a Ford. Under these conditions, S
has a justified true belief that r, knows his evidence for r to be
true, but does not know that r.
This, of course, is a variant of the familiar "Lucky Mr. Nogot"
counterexamplethat has proven especially intractablefor attempts to
specify the logically sufficientconditions for one's having knowledge.7
Shope's diagnosis suggests (p. 216) that by constructinga JEC for
the present sort of counterexample, we would expose a falsehood
94 PAUL K. MOSER
when attempting to explain what justifies the proposition that p:

Jones owns a Ford. On one variant of the familiar Lucky Nogot
counterexample,Jones is shamming, and so the relevant falsehood is
readily available: the false proposition that Jones' giving S the
relevant evidence stems from an intention to convey true informa-
tion. But this falsehood is clearly not presupposed by example (1),
concerning which it is perfectly coherent to suppose that Jones does
intend to convey true information while in the state of hypnosis.
Where, then, is the relevant falsehood that is to be exposed by con-
structinga JEC for (I)?The following section shows that this question
leads to trouble for Shope's diagnosis.
II
Suppose that we begin to develop the JEC for (I) as follows:

(E1) The proposition that q [= There is someone, whom S has
always found to be reliable and whom S has no good reason
to distrustat present, who has told S, his office-mate,that he
owns a Ford] is justified for S; and that fact justifies for S the
propositionthat r [= Someone in the office owns a Ford].
(E2) The proposition that m [=Jones, whom S has always found to
be reliable and whom S has no good reason to distrust at
present, has told S, his office-mate, that p: He, Jones, owns a
Ford] is justified for S; the proposition that m entails the
proposition that q; S recognizes that this entailment relation
holds; and all this makes (El) justified for S.
(E3) S's memory beliefs about what he has heard are, with very
little if any exception, veridical;S has the memory belief that
Jones has told him that p; S has no good reason to suppose
that the former memory belief is unveridical; S has good
reason to hold that justification is transmissible through
recognized entailment relations; and all this makes (E3)
justified for S.
Consistent with Shope's diagnosis, the above talk of a proposition's
(h's) being justifiedfor S is not intended to imply that S believes that
h; rather,it means simply that relative to all the evidence available to
S, h is justified. (Notice that such propositional justification might

also be relativized, without any difficulty for the present JEC, to an
epistemic community.) Now if (El)-(E3) are the initial links of an
appropriateJEC for example (I), it is highly doubtful that any false-
hoods will be exposed by the relevant JEC. Of course it might be
objected that the JEC for (I) must explain what justifies the proposi-
tion that Jones owns a Ford, and in doing so would expose a false
proposition. But it is not clear why the JEC explaining the justifica-
tion of r in (I) should be requiredto explain the justification of that
proposition, especially since that proposition is not a member of the
set of propositions constituting S's justification for r. The JEC
beginning with (El)-(E3) indicates that we can explain the justifica-
tion of r without explaining the justification of the proposition that
Jones owns a Ford. Thus, it is doubtful that the JEC for r in (I) will
expose a falsehood on which the justification of the latter proposition
apparently depends. But if this is so, Shope's diagnosis evidently has
the implausible implication that S knows that r in (I). In any case, the
present objection raises the crucial issue of when a proposition must
be included in a JEC explaining the justification of another proposi-
tion.
We can anticipate Shope's reply to the present objection, based on
the JEC (El)-(E3), by consideringhis view that S's knowledge that h
requiresS's belief that h to be justified "throughits connections with
a chain of propositions which themselves are justified in the sense of
being ones it would more manifest the rationality of members of the
epistemic community to accept in place of competing propositions
[and in place of withholding acceptance] when pursuing epistemic
goals" [e.g., goals like avoiding a false explanandum or explanans] (p.
138; cf. p. 217). The present notion of ajustifiedproposition is direct-
ly relevant to the objection at hand, inasmuch as it leads Shope to
reply in a specific way. Let us note, in this connection, that with
respect to the variant of the familiar Lucky Nogot counterexample
where Jones is shamming (which should not be confused with
example (I)), Shope suggests(pp. 216-17, and in correspondence)that
the presence or absence of the intention of the speakerto report true
informationmust be describedin the relevant JEC, since the issue of
the presence of such an intention is central to the rationality of an
96 PAUL K. MOSER
epistemic community considering whether to accept the speaker's

report.More specifically,Shope suggeststhat the rationalityof a com-
munity accepting the report would be undercut by the fact that the
reportdoes not stem from an intention to reporttrue information.On
this ground, Shope proposes that in the variant of the familiar Lucky
Nogot counterexample where Jones is shamming the deceptively
reportedproposition that Jones owns a Ford is not justified, and con-
sequently, that the inferred proposition that someone in the office
owns a Ford is not justified either.
Can a similar strategy, regardlessof its ultimate effectiveness, be
applied to counterexample (I), which does not depict Jones as
shamming? Shope's treatment of the Gettier-style counterexample
involving "Tom Grabit'sactual twin" (pp. 223-24) indicates that he
would apply a similar strategy.In that counterexample,a person, S,
reports,on the basis of visual experience, that his acquaintance,Tom
Grabit, stole a book from the library even though S is unable to
detect the differencebetween Tom's stealing the book and his actual
twin's doing so. Shope claims that S's lack of discriminative ability
here preventsit from being more rationalthan not for the membersof
an epistemic community to accept S's report on the basis of S's
testimony. Similarly, Shope would claim that in counterexample (I)
Jones' being in a state of hypnosis would prevent it from being more
rational than not for an epistemic community to accept Jones' report
that he owns a Ford on the basis of his testimony. And Shope would
claim, furthermore,that the proposition that Jones owns a Ford is
therefore not a justified proposition, in the envisaged circumstances,
and that this accounts for the lack of justificationand thus knowledge,
in (I), of the proposition that someone in the office owns a Ford.
However, two straightforwardconsiderationscount against Shope's
anticipated diagnosis of counterexample (I). First, it is quite
implausible to assume that the mere fact that Jones is in a state of
hypnosis prevents it from being more rational than not for an
epistemic community to accept Jones' report on the basis of his
testimony. For we can easily imagine a case where all the evidence
available to the members of an epistemic community either fails to
indicate that Jones is in a state of hypnosis, or more strongly,
indicates that Jones is a reliable reporterand thus is not in a state of
hypnosis. In such a case, clearly, the mere fact that Jones is in a state
of hypnosis would not prevent it from being more rational than not
for the membersto believe Jones' reporton the basis of his testimony.
(For similar reasons, it is implausible to suppose, with respect to the
counterexample where Jones is shamming, that the mere fact that
Jones does not intend to report true information prevents it from
being more rational than not for the members to believe his report.)
But if this is so, Shope's diagnosis fails to explain, in either case, why
the proposition reportedby Jones is actually unjustifiedfor the mem-
bers (including the experts) of the relevant epistemic community or
for S. And given this failure, the diagnosis also fails to explain why
the key inferredproposition that someone in the office owns a Ford is
unjustified.Certainly, under the envisaged circumstances,there is no
reason to think that the proposition in question would be unjustified
for the members of the actual epistemic community. Thus, it should
be noticed that for Shope's diagnosisto succeed, one would apparent-
ly have to claim that the justified proposition, q, in (El) and (I), does
not suffice to justify r, and that some false proposition is needed for
the justification of r. But it is highly doubtful that the justified q fails
to justify r. And this means that we evidently can constructa genuine
JEC like (El)-(E3) for r in (I) that does not expose a knowledge-
precludingfalsehood.
The second relevant considerationis that it still is not clear why a
JEC for r in (I) must explain the justification of the proposition that
Jones owns a Ford. Notice, in this connection, that the JEC (El)-(E3)
satisfies Shope's aforementioned requirement that propositional
knowledge that h must be justified through its connections with a
chain of propositions that it would be more rational than not for the
members of the epistemic community to accept. It is perfectly
compatible with example (I) to assume that each proposition in
(El)-(E3) is justified for the relevant epistemic community. But this
means that, on Shope's diagnosis, we are committed to the implaus-
ible view that S knows that r in (I).
III
The considerationsof Part II lead to the conclusion that Shope's diag-
98 PAUL K. MOSER
nosis fails to account for the lack of knowledge in Gettier-style

counterexampleslike (I). The major problem for the diagnosis is, of
course, that its use of a JEC does not guarantee that a knowledge-
precluding falsehood will be exposed in such Gettier-style counter-
examples. Granted, it would be false to assume, with respect to (I),
that Jones is not in a state of hypnosis. But this observationdoes not
explain how the presence of falsehood in (I) precludesknowledge;for
it does not explain how falsehood figuresin the relevantjustification.
Specifically, Shope's diagnosis fails to explain why the proposition q
in (I) does not justify r. What we need from a diagnosis of Gettier-
style counterexamples like Shope's is an explanation of the exact
roles falsehoods play in those examples. Such an explanation will
specify in general why the presence of falsehoods precludes knowl-
edge in certain cases and not in others. But if the above considera-
tions are correct, we cannot rely solely on the notion of a JEC to
provide the needed explanation. It follows further,then, that so far as
Shope's diagnosis goes, the Gettier problem has not actually been
solved.
Notice also that we cannot salvage Shope's diagnosis with the
likely proposal that for a person to have knowledge, there must be a
JEC that contains no falsehood and justifies no falsehood. In the JEC
for (I), the justified proposition, m, in (E2) evidentlyjustifies the false
propositionthat Jones is not claimingFord ownershipbecausehe is in
a state of hypnosis; given the justified m, one evidently could justifi-
ably believe that Jones is now a reliable reporter concerning his
alleged Ford ownership,and thus that he is not in a state of hypnosis.
Yet the proposal at hand is ultimately too demanding. It precludes,
for instance, one's knowing the following propositions:
Jones claims he owns a Ford;
Jones has always been trustworthyin the past;
Jones has often been seen drivinga Ford;
Jones has a Ford in his garage,
whenever the justifying evidence for those propositions justifies the
false proposition that Jones owns a Ford. In the latter case, the JEC
for the aforementionedfour propositions would justify a false propo-
sition, and so, given the present proposal, would preclude knowledge
of those propositions. But this is implausible. Of course the present

proposal might be revised to allow that a certain sort of JEC that
justifies a falsehood does not preclude knowledge. Yet it is not clear
what sort of JEC qualifies as such an exception.8 The following
section makes a more promising effort to analyze the notion of
propositional knowledge by means of the notion of epistemic
explanation.
IV
Let us consider whether we can undercutcounterexamplessuch as (I)

with the following initially plausible requirement:
ER. For S to have knowledge that p, there must be an
epistemic explanation of p that explains, solely by means
of true propositions, why S is justified in believing that p
on evidence, e, even if any other true proposition is con-
joined with e.
The basic assumption underlying ER is that the kind of evidence
appropriateto knowledge admits of an epistemic explanation that is
not underminedby the addition of any furthertrue propositions. An
epistemic explanation appropriate to knowledge, we might say, is
resistant to any truth, including any truth that is not part of the
knower's actual evidence base. Equivalently, we might say that the
kind of evidence requiredby knowledgeis truth-resistantin the sense
that its justificatory value is not underminedby the addition of any
truth.9The failure to acknowledgethis featureof knowledge,we have
seen in Part II, leaves an analysis of knowledgewide open to counter-
examples such as (I). Although ER seems to be sufficiently
demanding to explain the lack of knowledge in various Gettier-style
counterexamples, our first question should be whether it is too
demanding, whether there are cases of actual knowledge failing to
satisfy ER.
One pertinent example, in the present connection, resembles the
above-describedcounterexampleinvolving Tom Grabit'sactual twin,
except it involves Tom's demented mother, Mrs. Grabit:
100 PAUL K. MOSER
(II) S believes, on the basis of his visual experience, that his

acquaintance, Tom Grabit, stole a book from the library. But
Tom's twin is simply a fiction of the demented mind of Tom's
mother, Mrs. Grabit, who has claimed thatf Tom's twin was
at the libraryat the time of the theft, but Tom was not.10
The true proposition that Mrs. Grabit claimed that f is, accordingto
some philosophers, a defectiveunderminerof S's evidence, since if it
undermines,it does so only because it justifies S in believing the false
propositionft Accordingly, it might be claimed that such a defective
underminer does not prevent S from knowing that Tom stole the
book, even though the undermineris true and, when conjoined with
S's evidence, undermines S's justification for the belief that Tom
stole the book. However, this example involves a rather unclear
assumption. Once S notes that Mrs. Grabit, being demented, is
not a reliable source of information about her son's whereabouts,S
may infer that her claim that ft is not reliable and that thereforeher
claim, when added to S's evidence, does not make it improbablefor S
that Tom stole the book. But this just means that S may deny that
Mrs. Grabit's claim that f is an underminer in any relevant sense,
with respect to S's evidence. Thus, the proponent of ER apparently
need not deny that S has knowledgein the case in question.
However, we can make the present example more difficult for ER,
if we refrain from adding to S's evidence the true proposition that
Mrs. Grabit is demented. Suppose that it still is a fact that Mrs.
Grabit is demented, but that we add instead to S's evidence only the
true proposition that m: Mrs. Grabit claimed that Tom was nowhere
near the library,and mothers are typically highly reliable as a source
of informationabout their sons' general whereabouts.In this case, ER
implies that S does not know that Tom stole the book, since S's
evidence, when conjoined with the true proposition m, fails to justify
for S the proposition that Tom stole the book. But this is an incorrect
implication of ER, given the fact that Mrs. Grabit is demented. The
problemhere is due to the assumptionof ER that alltruths that under-
mine justificationpreclude knowledge.
The example (II) raisesthe question of when truthsthat undermine
justification also preclude knowledge. A simple likely answer to this
question, in the spirit of ER, is:

(A) A truth, t, that underminesS's evidence, e, for p precludes
S's knowledge that p on e only if there is no furthertruth,
t', that restores the justification of p when conjoined with
(e&t).
In the example (II) involving the demented Mrs. Grabit, there is a

furthertruth that, when conjoined with S's evidence e and the under-
miner m, restores S's justification for the belief that Tom stole the
book; the further truth, of course, is that Mrs. Grabit is demented.
Thus, it might be proposedthat ER be supplementedwith the require-
ment that the underminersof justification must themselves be truth-
resistantin the sense specifiedby (A).
We then have the following emendationof ER:
ER*. For S to have knowledge that p on justifying evidence e, e
must be such that for every true proposition t that, when
conjoined with e, undermines S's justification for p on e,
there is a true proposition t' that restoresthe justification
of p for S on (e & t).
Since it is not obvious that there are cases of actual knowledge that
fail to satisfy the requirementset by ER*, let us postpone the ques-
tion whether ER* is too demanding. Let us ask instead whether the
requirementset by ER* can plausibly be taken as stating a sufficient
condition for evidence that satisfies the fourth condition for knowl-
edge. I believe we can illustratethat ER* does not state such a suffi-
cient condition by considering one of Gettier's original counter-
examples:
(III) Smith is justified in believing the false proposition that (i)
Jones owns a Ford. On the basis of (i) Smith infers, and thus is
justified in believing, that (ii) either Jones owns a Ford or
Brown is in Barcelona. However, as it turns out, Brown
happens to be in Barcelona,and so (ii) is true. Thus, although
Smith is justified in believing the true proposition (ii), Smith
does not know (ii).
Here we have an example where the requirementset by ER* is satis-
102 PAUL K. MOSER
fied, but where S lacks knowledgewhile havingjustified true belief In

such an example, S is justified in believing a false proposition,f, on
evidence e, and thus, via deductive inference, is justified in believing
a disjunction, (for b), where b is true. Further,when conjoined with
S's evidence, the true proposition-fundermines S's justificationfor (f
or b). But there is a true proposition that restoresthis justificationfor
S; this is the true proposition b. Thus ER*, when construedas stating
a sufficient condition for the fourth condition for knowledge, implies
that S knows that (for b); but this, of course, is an incorrectimplica-
tion of ER* thus construed.
A notion of truth-resistantevidence that avoids the difficulty for
ER* posed by example (III)is the following:
TR. S's justifying evidence e for p is truth-resistantif and only
if for every true proposition t that, when conjoined with e,
underminesS's justificationforp on e, there is a true prop-
osition, t', that restores the justification of p for S, on
e & t & t', in a way that S is actuallyjustified in believingp
to be true.
The notion in TR of a way that S is actuallyjustiJiedin believingp to
be true presupposesthe notion of an alethically variableproposition.
Let us say that a proposition, p, is alethically variable if and only if
there is more than one way for p to be true. Nonredundantdisjunc-
tions and existential generalizationsare paradigmcases, but not the
only cases, of alethically variable propositions. A way for a disjunc-
tion to be true, as we know from the familiartruth tables, is described
by any set of its disjuncts;and a way for an existential generalization
to be true is described by any set of its instantiations. Thus, with
respect to the counterexample (III), a way for the disjunction that
(Jones owns a Ford or Brown is in Barcelona)to be true is described
by the true proposition that Brown is in Barcelona,and another way
is describedby the false proposition that Jones owns a Ford. Given a
plurality of propositionally distinct disjuncts and instantiations,
alethically variable propositions of the present sort qualify as propo-
sitions for which there are various ways of being true.
The present notion of an alethically variable proposition presup-
poses a notion of propositionalidentity. Following Chisholm, we may
rely on the following criterion of propositional identity: A proposi-

tion, p, is identical with a proposition, q, if and only if p and q are
necessarily such that (i) p is true if and only if q is true, and (ii) any
person accepts p if and only if he accepts q. 1 Given this criterion, we
can make sense of the notion of a proposition'sbeing true in one way
ratherthan another, since ways of being true arejust propositionssuf-
ficient for the truth of some proposition. Also given this criterion,we
can see that a redundant disjunction of the form 'p or p' is not
genuinely an alethically variableproposition.
Returningthen to principle TR, we may understandthe notion of
a way that S is actuallyjustified in believing p to be true, relative to a
disjunctionor an existential generalization,as follows:
For any disjunction, (p or q), a way that S is actually
justified in believing (p or q) to be true is describedby p if
and only if S's justifying evidence for (p or q) consists, at
least in part, of justifying evidence for p.
This explication applies to alethically variable existential generaliza-
tions as well as disjunctions, since such generalizations logically
behave as disjunctions.Also, this explication applies to cases, such as
the familiar lottery situations, where a person has justifying evidence
for a disjunction, but lacks justifying evidence concerning which dis-
junct is true.12 In such cases, a way that the person is actuallyjustified
in believing the disjunctionto be true is describedby the disjunction
itself, and not by any proper subset of its disjuncts. This is just to say
that p in the above explication may itself be disjunctive.
However, we should not assume that only disjunctionsand existen-
tial generalizationsqualify as alethically variable propositions. There
are various ways for a universal proposition like 'All epistemologists
are poor' to be true. One such way is described by the statement
'Jones is the only epistemologistand Jones is poor', and another such
way is described by the statement 'Smith is the only epistemologist
and Smith is poor'. Thus, Gettier-stylecases can arise with universal
propositions as well as with disjunctive and existential propositions.
And, therefore,the notion in TR of a way that S is actually justified
in believing p to be true should not be restrictedto disjunctionsand
existential generalizations.Notice also that TR allows for cases where
104 PAUL K. MOSER
there is more than one way that a person is justified in believing a

proposition to be true. It is plausible to suppose that for there to be
knowledge in such cases, a person'sjustifying evidence must be truth-
resistantwith respect to at least one of the ways that he is justified in
believing the relevantproposition to be true.
We now need to consider various Gettier-stylecounterexamplesas
a means of testing the analysis of knowledge that includes a fourth
condition of evidential truth-resistanceas explicated by TR. Let us
take the relevant analysis of propositional knowledge to be the fol-
lowing:
PK. A person, S, has propositional knowledge that p if and
only if: p is true; S has justifying evidence, e, for p that is
truth-resistantin the sense specified by TR; and S believes
that p on the basis of e.
The key thesis of PK may be regarded,for purposes of simplicity, as
the requirementthat propositionalknowledgemust have justifiedtrue
belief sustained by the totalityof truths.
Considering first the above-described example (III), we should
recall that in that example Smith is justified in believing the false
proposition:
(i) Jones owns a Ford,
which, according to (III), justifies the following true disjunction for
Smith:
(ii) EitherJones owns a Ford or Brownis in Barcelona.
But Smith does not know (ii). Once Smith's evidence base is
expanded by the true proposition that Jones does not own a Ford, we
see that Smith's lack of knowledge in example (III) is explainable by
PK; for then we see that Smith is not justified in believing that (i).
The explanation of this undermining of justification is simple:
Smith's initial evidence for (i), when supplementedby the true propo-
sition that Jones does not own a Ford, does not make (i) more
probable than not. A basic requirement for the justification of any
proposition, p, is that p be more probablethan its denial. Thus, if on
the expanded evidence Smith is justified in believing that Jones does
not own a Ford, he is not, on that evidence, justified in believing that

Jones owns a Ford. So, the justification for the latter proposition has
been underminedby a true proposition.
Furthermore,given PK, there is not in (III)the appropriatekind of
restorerof the justification underminedby the true proposition that
Jones does not own a Ford. One inadequaterestoreris the true propo-
sition that Brown is in Barcelona. Given PK, we can explain the
inadequacy of the restoration of justification provided by the latter
proposition, on the groundthat the restorationhas nothing to do with
S's actual justification for the disjunction (Jones owns a Ford or
Brown is in Barcelona). S's actual justification consists solely of
evidence for the false disjunct that Jones owns a Ford; it does not
include justifying evidence for the disjunct that Brown is in
Barcelona. Thus a restoration of S's undermined justification that
supports only the latter disjunct is, given PK, inadequate to sustain
knowledge for S. For in such a case, it is not S's actual evidence
relative to the false disjunct that is truth-resistant.Consequently, PK
explains S's lack of knowledgein example (III).
Considering next the troublesome counterexample (I), we should
recall that in that example S knows the following proposition to be
true:
(i) Mr. Jones, whom S has always found to be reliable and
whom S has no good reason to distrustat present, has told
S, his office-matethat he, Jones, owns a Ford.
Further, according to (I), S correctly deduces from (i), and thus is
justified in believing, the following true existential generalization:
(ii) There is someone, whom S has always found to be reliable
and whom S has no good reason to distrustat present,who
has told S, his office-mate,that he owns a Ford.
On the basis of (ii), then, S infers and is justified in believing the fol-
lowing true proposition:
(iii) Someone in the office owns a Ford.
However, S does not know (iii), due to the following condition:
106 PAUL K. MOSER
(c) Jones has told S that he owns a Ford only because of the
hypnotic state he is currentlyin.
PK accounts for S's lack of knowledge in the present case, because
when we add the true proposition(c) to S's evidence, S is not justified
in believing (iii) on the basis of that expanded evidence; for that
evidence does not make (iii) more probable than not. The reason it
does not make (iii) more probable than not is simple: that expanded
evidence undercuts the presumed reliability of Jones' claim that he
owns a Ford. Furthermore,there is in example (I) no truth providing
an adequate restoration of S's underminedjustification. Therefore,
S's evidence is not truth-resistantin the sense requiredby PK, and so
PK accounts for S's lack of knowledgein example (I).
PK also explains the lack of knowledge in the aforementioned
more familiar "Lucky Mr. Nogot" counterexample where Jones is
shamming. For when we expand S's evidence to include the true
proposition that Jones is shamming, that expanded evidence base
justifies for S neither the proposition that someone who is currentlya
reliable office-mate has said that he owns a Ford nor the proposition
that someone in the office owns a Ford. Further,there is not in that
example the sort of restorer of justification required by PK. The
above-mentioned considerationsexplaining the lack of knowledge in
example (I) apply here as well.
PK is equally effective with regardto the following example due to
Gettier:
(IV) Smith and Jones have applied for the same job. Smith is
justified in believing that (i) Jones will get the job, and that (ii)
Jones has ten coins in his pocket. On the basis of (i) and (ii)
Smith infers, and thus is justified in believing, that (iii) the
person who will get the job has ten coins in his pocket. How-
ever, as it turns out, Smith himself will actually get the job,
and he also happens to have ten coins in his pocket. Thus,
although Smith is justified in believing the true proposition
(iii), Smith does not know (iii).
In this example Smith is justified in believing the false proposition:
(i) Jones will get the job.
and the proposition:

(ii) Jones has ten coins in his pocket.
According to (IV), when conjoined these propositions justify for
Smith the following true proposition:
(iii) The person who will get the job has ten coins in his
pocket.
However, there is in (IV) a true proposition that clearly undermines
Smith's justification for (iii) as well as (i) when that proposition is
added to his justification. The proposition, of course, is that Smith
will get the job. When the latter proposition is conjoined with Smith's
evidence, it is not probable for Smith that Jones will get the job; for
the true proposition that Smith will get the job implies, of course,
that Jones will not get the job. Moreover,there is in (IV) no adequate
restorer of Smith's justification for (iii); for there is no adequate
restorerof Smith's actualjustificationfor the belief that Jones will get
the job. Thus, Smith's evidence in (IV) is not truth-resistantin the
sense required by PK; and, therefore, PK accounts for the lack of
knowledgein (IV).
It is also useful to mention the relevance of PK to the earlier-
described counterexample involving Tom Grabit's actual twin: S
believes, on the basis of his visual experience, that his acquaintance,
Tom Grabit, stole a book from the library;but, unknown to S, Tom
has an identical twin brotherwho was in the libraryduring the time
of the theft. Given PK, we can deny that S has knowledge in this
example. For when we add to Ss evidence the true propositionthat S
is unable to detect the difference between Tom's stealing the book
and his twin's doing so, that evidence fails to justify for S the proposi-
tion that Tom stole the book. Furthermore,there is not in this exam-
ple the appropriatesort of restorerof S's justification.Thus, since S's
justification is not truth-resistantin the manner requiredby PK, the
presentexample is not troublesomefor PK.
It might be argued,more generally,that PK can explain the lack of
knowledge in every Gettier-stylecounterexample,on the ground that
the characteristicfeatureof such a counterexampleis that it identifies
a case of justified true belief were the justification is not truth-
108 PAUL K. MOSER
resistantin the manner requiredby PK. In any such case, there is not
available an epistemic explanation of the relevantjustified true belief
that consists solely of true propositions and is truth-resistantin the
required manner. Given PK, then, Gettier-style counterexamples
should cease to perplex us.
However, we must ask whether PK is too demanding, whether
there are cases of propositional knowledge that fail to satisfy the
requirementof truth-resistantjustificationset by PK. This leads us to
reconsiderthe earlier-mentionedexample (II)that raisedproblemsfor
principle ER. In that example, S believes, on the basis of his visual
experience, that his acquaintance,Tom Grabit, stole a book from the
library, but Tom's twin is just a fiction of the imagination of Tom's
demented mother, Mrs. Grabit, who has claimed that f: Tom's twin
was at the library at the time of the theft, but Tom was not. PK
accounts for S's having knowledge in this example, on the ground
that S's justification is truth-resistantin the requiredway. When we
add to S's evidence the truth that Mrs. Grabit's claim that f is mis-
taken, we preclude her claim's serving as an underminer of S's
evidence. And this is true even of the more difficult variation on the
present example where we add to S's evidence only the potential
undermining truth that Mrs. Grabit has claimed that f and mothers
are typically highly reliable as a source of information about their
sons' general whereabouts. Consequently, the example that was
troublesomefor ER is easily handledby PK.
Another example that might be used to arguethat PK is too strong
also resembles the counterexample involving Tom Grabit's actual
twin, except it includes the following situation: while observing
Tom's theft, S momentarilyentertainsthe proposition that Tom has
an identical twin who was in the libraryduring the time of the theft;
but S also knows that he does not actually believe that proposition.13
In this example, it might be argued, the momentarily entertained
proposition - call it d - is, if true, a genuine underminerof S's justifi-
cation for the belief that b: he does not believe d; for if d is added to
S's evidence for believing that b (in the sense that S believes that d),
then S is not justified in believing that b. Therefore, the argument
concludes, PK improperly leads us to deny that S knows that b. But
this argument rests on a confusion: a confusion of the proposition
that d and the proposition that S believesthatd. Although the latter

proposition is, if true, a genuine underminer of S's evidence for b,
the former clearly is not. Furthermore, when d is added to S's
evidence, in accordance with PK, only the proposition that d is
added; the proposition that S believes that d is not thereby added to
S's evidence, even if we therebyassumethat S believes that d. But, of
course, our assumingthe proposition that S believes that d (simply
for purposes of testing PK) neither makes this proposition true nor
makes it part of S's actual evidence. Notice also that one's addingthe
proposition that S believes that d to S's actual evidence for b is
definitely not condoned by the testing of PK, since that proposition,
being the contradictoryof the true b, is false. Consequently, there is
no reason to think that the presentcase is troublesomefor PK.
Still another case deserves brief consideration for purposes of
testing PK:
(V) Mr. Relator, a teacher whom S knows to be generally reliable
as a source of informationabout events at school, reportsto S
the informationthat led Mr. Relatorjustifiablyto believe that
q: Mr. Nogot owns a Ford. The relevant informationis p: Mr.
Nogot, one of my students, drove a Ford in front of me,
affirmedthat he owns it, showed me papers to that effect, and
has been generally reliable in past dealings with me.
Accordingly, S knows that p. Further, Mr. Relator and S are
justified in believing that q, as neither of them has any reason
to believe it is false. But, as it turns out, q is false.14
The question raised by (V) is whether PK improperly denies that S
knows that p in the specified situation where S does not believe that
-q. Specifically, does the addition of the true proposition that -q to
S's evidence undermine S's justification for p and thereby preclude
S's knowledge that p? The answer, it seems quite clear, is no. For
even if S justifiably assumes that Nogot does not own a Ford, it can
still be highly probable, and thus justified, on S's evidence that the
report of the generally reliable Mr. Relator is correct; that is, the
reportthat Nogot drove a Ford in front of Relator, claimed to own it,
displayed papersto that effect, and has been generally reliable in past
dealings with Relator. Indeed, the justification for p can still be truth-
110 PAUL K. MOSER
resistant in the manner required by PK. For -q neither is incom-

patible with p nor makes p improbable. Consequently, PK does not
preclude S's knowledge that p, and so is not shown to be excessively
demandingby example (V).
Another relevantexample challengingthe requirementof evidenti-
al truth-resistanceset by PK is the following:
(VI) S attends a wedding ceremony where two of his friends
become married. The ceremony is performed without any
errorsby the local Bishop. After witnessing the ceremony, S
knows that his friends are married. However, unknown to
anyone witnessing the ceremony, the local Cardinal goes
insane at the time of the ceremony, and falsely denounces the
Bishop as a fraud who is not authorized to perform
weddings.15
Clearly, this example is no more challengingto PK than is the afore-
mentioned case of the demented Mrs. Grabit. In fact, a direct
analogue of the above explanation of the demented Mrs. Grabit case
providedby PK applies, with equal effectiveness,to example (VI).
A final pair of examples that deservesconsiderationfor purposesof
testing PK is the following:
(VII) S has a justified true belief that his two friends were just
marriedby the local Bishop, but there is a true proposition,
t, which underminesSs justification when conjoined with it
(in the way required by PK), and which is such that it is
humanly impossible (i.e., impossible given human limita-
tions) for S to be justified in believing that t.
(VIII) S has a justified true belief that his two friends were just
marriedby the local Bishop, but there is a true proposition,
t, which underminesS's justificationwhen conjoined with it
(in the way required by PK), and which is such that it is
physically impossible for S to be justified in believing that t.
It might be argued that examples (VII) and (VIII) are cases where S
has knowledge,and thus that PK is too strong. However, I believe the
more enlightening moral to those two examples is that we should
countenance varyingstrengthsof knowledge, strengthsdeterminedby
accessibility qualifications placed on the set of appropriate true

underminers.Given this moral, we should countenance a variety of
concepts of knowledge weaker than PK, concepts formulated by
means of various accessibility-to-underminersqualificationsadded to
PK.
PK is our most demanding concept of propositional knowledge,
since it assumes that it need only be logically possible for a person to
be justified in believing the appropriateunderminers.This is a viable
concept of knowledge, at least inasmuch as many philosophers are
evidently inclined to endorse it as the concept of knowledge. But it
seems to me gratuitousto deny that there are concepts of knowledge
less strenuous than PK. One weaker concept of knowledge is
suggested by example (VIII): according to this concept, true under-
miners of the sort mentioned in PK must be such that it is physically
possible for a person to be justified in believing them. And a still
weaker concept is suggestedby example (VII):accordingto this con-
cept, true underminersmust be such that it is humanly possible for a
person to be justified in believing them. (Here I assume that not all
physical possibilities are human possibilities.)It appearsthat different
philosophers have presupposedeach of these three viable concepts of
propositionalknowledge.
The difference between the present concepts, quite clearly, is due
to differingdegrees of accessibility-to-underminersinvolved with the
key notion of truth-resistantevidence. The availability of these dif-
fering degrees of accessibility should lead us to countenance various
concepts of propositional knowledge stemming from PK. Some of
these concepts will appear to be too demanding relative to certain
Gettier-style counterexamples;whereas others will appear to be too
weak. However, so long as these concepts build on the truth-
resistancerequirementset by PK, and differonly on the accessibility-
to-underminersqualification, they will survive the range of Gettier-
style examples where our epistemological intuitions are quite clear.
The remaining Gettier-style examples are subject to divergent treat-
ments, since they constitute those cases where our intuitions waver
and thus where it is appropriate to entertain divergent concepts of
propositional knowledge. Thus, although PK does not have a
monopoly on the concept of propositionalknowledge,it provides the
112 PAUL K. MOSER
essential core for that concept. In doing so, PK takes the mystery out
of the Gettier-style counterexamples that intuitively demand the
formulation of evidential constraints on propositional knowledge.
The essential constraintprovidedby PK may be characterizedsimply
as justification-sustenanceby the totality of truths.
v
In summary, then, we have seen that the analysis of propositional
knowledge provided by PK explains not only Gettier's original two
counterexamples, but also some of the most difficult Gettier-style
counterexamples, e.g., example (I) above, the familiar Lucky Mr.
Nogot case, and the cases involving Tom Grabit and his demented
mother. We have also seen how this analysis might be understoodas
improvingsubstantiallyon the use of epistemic explanation in Robert
Shope's initially promising analysis, and how it can meet several
likely objections to the effect that it is too demanding. Perhaps it is
worth emphasizing also that this analysis, unlike typical defeasibility
analyses, does not employ a subjunctive conditional, and therefore
does not fall prey to the objection that it commits the conditional
fallacy.'6 The analysis can be put simply in standardquantificational
form as follows: For every proposition, p, and every instance of
evidence, e, S knows that p on e if and only if p is true; S has justify-
ing evidence, e, for p that is truth-resistantin the sense specified by
the (subjunctive-free)principle TR above; and S believes that p on
the basis of e. Stated thus, it should also be emphasized, the analysis
does not run afoul of the principle from probability theory that the
probabilityof a non-redundantconjunction is less than the probabili-
ty of its individual conjuncts having a probabilityless than 1; for the
analysis proceeds on the assumption that any true proposition, t, is
added to S's evidence, e, in the sense that S is justified in believing
that (e & t), and thus the relevant conjunction is not to be taken as
unjustifiedfor S due to low probability.Another virtue of the present
analysis, moreover, is that it clearly allows for non-deductivejustify-
ing evidence as a component of propositionalknowledge. Finally, the
analysis provided by PK enables us to see how to generate less
strenuous, but equally plausible, concepts of propositional knowl-
edge, particularly in connection with Gettier-style examples where

our epistemological intuitions waver. Given these considerations,we
may reasonably endorse PK as a most promising explication of the
basic concept of propositional knowledge. Or, more boldly, we may
conclude that it provides an analysis of propositional knowledge that
puts the Gettier problem to rest once and for all.'7
NOTES
See Gettier, 'Is Justified True Belief Knowledge?',Analysis 23 (1963), 121-23,

reprintedin EmpiricalKnowledge,ed. P. Moser (Totowa, NJ: Rowman & Littlefield,
1986).
2 Such a defeasibilityapproach,with distinctivevariations,is represented,for instance,
in Keith Lehrerand Thomas Paxson, 'Knowledge:UndefeatedJustifiedTrue Belief',
The Journal of Philosophy66 (1969), 225-37; MarshallSwain, 'EpistemicDefeasibili-
ty', AmericanPhilosophicalQuarterly11 (1974), 15-25; and Peter Klein, 'Knowledge,
Causality, and Defeasibility', The Journal of Philosophy 73 (1976), 792-812. For
decisive objections to this approach, see Robert Shope, The Analysis of Knowing
(Princeton:PrincetonUniversityPress, 1983), Chapter2.
3 Such a view is suggested,even if not explicitly endorsed,in Robert G. Meyersand
Kenneth Stern, 'KnowledgeWithout Paradox', The Journal of Philosophy70 (1973),
147-60; David M. Armstrong,Belief, Truth,and Knowledge(Cambridge:Cambridge
University Press, 1973), p. 152; and Gilbert Harman, Thought(Princeton:Princeton
UniversityPress, 1973), p. 47. This kind of view is challengedin RichardFeldman,'An
Alleged Defect in Gettier Counter-Examples',AustralasianJournal of Philosophy52
(1974), 68-69, reprintedin EmpiricalKnowledge,ed. P. Moser (Totowa, NJ: Rowman
& Littlefield,1986).
4 This approach is developed by Robert Shope in 'Knowledge and Falsity',
PhilosophicalStudies 36 (1979), 389-405, and in TheAnalysis of Knowing,Chapter7.
It derivesin part from the analysisof knowledgein ErnestSosa, 'How Do You Know?',
AmericanPhilosophicalQuarterly11 (1974), 113-22.
5 These and all subsequentparentheticalpage-numbersreferto Shope's The Analysis
of Knowing.
6 This counterexampleis inspired in part by Richard Feldman's example in 'An
Alleged Defect in Gettier Counter-Examples',AustralasianJournal of Philosophy52
(1974), 68-69. Cf. Keith Lehrer,'The GettierProblemand the Analysisof Knowledge',
in Justificationand Knowledge,ed. G.S. Pappas(Dordrecht:D. Reidel, 1979), p. 75.
7 A simple version of the Lucky Nogot counterexampleis discussed in Shope, The
Analysisof Knowing,pp. 68-69.
8 For substantiationof the presentpoint, see Shope, TheAnalysisof Knowing,Chapter
4, wherethe relevantviews of R. M. Chisholmand ErnestSosa are criticallydiscussed.
9 Here let us make the natural assumption that whenever a person, S, has truth-
resistantjustifying evidence for a proposition,p, there will be a correspondingtruth-
resistantepistemic explanation of p for S (which consists solely of true propositions),
and conversely.
10 This example is due to Keith Lehrerand Thomas Paxson, 'Knowledge:Undefeated
JustifiedTrue Belief', TheJournalofPhilosophy66 (1969), 228.
Chisholm suggeststhis criterion in Person and Object(La Salle, IL: Open Court,
1976), pp. 118-24. For an effort to challenge W. V. Quine's notorious qualms about
114 PAUL K. MOSER
criteria of propositionalidentity, see my 'Types, Tokens, and Propositions:Quine's

Alternative to Propositions',Philosophy and PhenomenologicalResearch 44 (1984),
361-75.
12 For discussion of the epistemological significanceof such lottery cases, see Paul
Moser, Empirical Justification (Boston: D. Reidel, 1985), Chapter 1, and Henry
Kyburg, Epistemology and Inference (Minneapolis:University of Minnesota Press,
1983), Chapters14 and 15.
13 This examplecomes froni Shope, TheAnalysisof Knowing,p. 71, whereit is used to
oppose Peter Klein's defeasibilityanalysisin 'MisleadingEvidenceand the Restoration
of Justification',PhilosophicalStudies37 (1980), 81-89.
14 The presentexamplecomes from Shope, TheAnalysisof Knowing,p. 62.
15 This example comes from Marshall Swain, 'Epistemic Defeasibility', American
PhilosophicalQuarterly11 (1974), 18, reprintedin Essays on Knowledgeand Justifica-
tion, eds. G. S. Pappasand M. Swain(Ithaca:CornellUniversityPress, 1978), p. 165.
16 For the details on this fallacy, see Robert Shope, 'The ConditionalFallacy in Con-
temporaryPhilosophy', TheJournalof Philosophy75 (1978), 397-413. In Chapter2 of
The Analysis of Knowing,Shope illustrateshow this fallacy is frequentlycommittedby
defeasibilitytheorists.
17 Various sections of this article have benefitedfrom comments from Robert Audi,
RobertShope, HarryGensler,Al Mele, and ArnoldvanderNat.
Departmentof Philosophy,
Loyola Universityof Chicago,
Chicago,IL 60626,
U.S.A.
View publication stats

Moser Proposition Al Knowledge Philosophical Studies 1987

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Moser Proposition Al Knowledge Philosophical Studies 1987

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Article in Philosophical Studies · January 1987

The user has requested enhancement of the downloaded file.

(Received 18 August, 1986)

Propositional knowledge is knowledge that some proposition is true.

PhilosophicalStudies52 (1987) 91-114.

modification is relatively new, and so will be considered, but found

Robert Shope's recent analysis of propositional knowledge relies on

proposition, p, is just a true conjunction of the form: '1f and that

when attempting to explain what justifies the proposition that p:

Suppose that we begin to develop the JEC for (I) as follows:

S, h is justified. (Notice that such propositional justification might

epistemic community considering whether to accept the speaker's

The considerationsof Part II lead to the conclusion that Shope's diag-

nosis fails to account for the lack of knowledge in Gettier-style

of those propositions. But this is implausible. Of course the present

Let us consider whether we can undercutcounterexamplessuch as (I)

(II) S believes, on the basis of his visual experience, that his

question, in the spirit of ER, is:

In the example (II) involving the demented Mrs. Grabit, there is a

fied, but where S lacks knowledgewhile havingjustified true belief In

rely on the following criterion of propositional identity: A proposi-

there is more than one way that a person is justified in believing a

not own a Ford, he is not, on that evidence, justified in believing that

and the proposition:

that d and the proposition that S believesthatd. Although the latter

resistant in the manner required by PK. For -q neither is incom-

accessibility qualifications placed on the set of appropriate true

edge, particularly in connection with Gettier-style examples where

See Gettier, 'Is Justified True Belief Knowledge?',Analysis 23 (1963), 121-23,

criteria of propositionalidentity, see my 'Types, Tokens, and Propositions:Quine's

You might also like